Income Prediction With Census Data

Census Income Data obtained from UC Irvine Machine Learning Repository
Link: https://archive.ics.uci.edu/dataset/20/census+income

Github repository: https://github.com/wei-hei-nip/ML-Income-Prediction.git

1. Preparing the Data

# Import libraries needed
import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
import seaborn as sns
from sklearn.model_selection import train_test_split
from sklearn.metrics import classification_report,confusion_matrix, ConfusionMatrixDisplay

# Load data
column_title = ['age','workclass','fnlwgt','education','education-num','marital-status','occupation', 'relationship', 'race', 'sex','capital-gain','capital-loss','hours-per-week', 'native-country', 'income']
df1 = pd.read_csv('data/adult.data', header=None, names=column_title)
df2 = pd.read_csv('data/adult.test', header=None, names=column_title)
df_census = pd.concat([df1, df2],axis=0)

# Info
df_census.info()

<class 'pandas.core.frame.DataFrame'>
Index: 48843 entries, 0 to 16281
Data columns (total 15 columns):
 #   Column          Non-Null Count  Dtype  
---  ------          --------------  -----  
 0   age             48843 non-null  object 
 1   workclass       48842 non-null  object 
 2   fnlwgt          48842 non-null  float64
 3   education       48842 non-null  object 
 4   education-num   48842 non-null  float64
 5   marital-status  48842 non-null  object 
 6   occupation      48842 non-null  object 
 7   relationship    48842 non-null  object 
 8   race            48842 non-null  object 
 9   sex             48842 non-null  object 
 10  capital-gain    48842 non-null  float64
 11  capital-loss    48842 non-null  float64
 12  hours-per-week  48842 non-null  float64
 13  native-country  48842 non-null  object 
 14  income          48842 non-null  object 
dtypes: float64(5), object(10)
memory usage: 6.0+ MB

# Futher check empty value
df_census = df_census.replace(' ?', np.NaN) # Some column found unknown value "?"
df_census[df_census.isnull().any(axis=1)]

	age	workclass	fnlwgt	education	education-num	marital-status	occupation	relationship	race	sex	capital-gain	capital-loss	hours-per-week	native-country	income
14	40	Private	121772.0	Assoc-voc	11.0	Married-civ-spouse	Craft-repair	Husband	Asian-Pac-Islander	Male	0.0	0.0	40.0	NaN	>50K
27	54	NaN	180211.0	Some-college	10.0	Married-civ-spouse	NaN	Husband	Asian-Pac-Islander	Male	0.0	0.0	60.0	South	>50K
38	31	Private	84154.0	Some-college	10.0	Married-civ-spouse	Sales	Husband	White	Male	0.0	0.0	38.0	NaN	>50K
51	18	Private	226956.0	HS-grad	9.0	Never-married	Other-service	Own-child	White	Female	0.0	0.0	30.0	NaN	<=50K
61	32	NaN	293936.0	7th-8th	4.0	Married-spouse-absent	NaN	Not-in-family	White	Male	0.0	0.0	40.0	NaN	<=50K
...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...
16209	21	NaN	212661.0	Some-college	10.0	Never-married	NaN	Own-child	White	Female	0.0	0.0	30.0	United-States	<=50K.
16240	73	NaN	144872.0	HS-grad	9.0	Married-civ-spouse	NaN	Husband	White	Male	0.0	0.0	25.0	Canada	<=50K.
16252	81	NaN	26711.0	Assoc-voc	11.0	Married-civ-spouse	NaN	Husband	White	Male	2936.0	0.0	20.0	United-States	<=50K.
16266	50	Local-gov	139347.0	Masters	14.0	Married-civ-spouse	Prof-specialty	Wife	White	Female	0.0	0.0	40.0	NaN	>50K.
16278	64	NaN	321403.0	HS-grad	9.0	Widowed	NaN	Other-relative	Black	Male	0.0	0.0	40.0	United-States	<=50K.

3621 rows × 15 columns

# Drop empty row
df_census = df_census.dropna(how='any',axis=0) 
df_census.shape

(45222, 15)

# Drop Working Class == "Without-pay"
df_census = df_census.drop(df_census[df_census.workclass == ' Without-pay'].index)
df_census.shape
df_census.index = range(len(df_census))
df_census.shape

(45190, 15)

# redefining datatype base on difference between df1.info() adn df2.info()
df_census = df_census.astype({"age":np.int64,"fnlwgt":np.int64, 'education-num':np.int64,'capital-gain':np.int64,'capital-loss':np.int64,'hours-per-week':np.int64, 'income':"string"}) 
print(df_census.info())

<class 'pandas.core.frame.DataFrame'>
RangeIndex: 45190 entries, 0 to 45189
Data columns (total 15 columns):
 #   Column          Non-Null Count  Dtype 
---  ------          --------------  ----- 
 0   age             45190 non-null  int64 
 1   workclass       45190 non-null  object
 2   fnlwgt          45190 non-null  int64 
 3   education       45190 non-null  object
 4   education-num   45190 non-null  int64 
 5   marital-status  45190 non-null  object
 6   occupation      45190 non-null  object
 7   relationship    45190 non-null  object
 8   race            45190 non-null  object
 9   sex             45190 non-null  object
 10  capital-gain    45190 non-null  int64 
 11  capital-loss    45190 non-null  int64 
 12  hours-per-week  45190 non-null  int64 
 13  native-country  45190 non-null  object
 14  income          45190 non-null  string
dtypes: int64(6), object(8), string(1)
memory usage: 5.2+ MB
None

# Drop extreme value in "capital-gain"
df_census = df_census[df_census["capital-gain"] < 99999]
df_census.shape

(44962, 15)

# Drop featureless column "fnlwgt"
df_census = df_census.drop("fnlwgt",axis='columns')
df_census.shape

(44962, 14)

# Count "income" label values
df_census = df_census.replace(' >50K.', ' >50K')
df_census = df_census.replace(' <=50K.', ' <=50K')
df_census["income"].value_counts()

income
<=50K    33986
>50K     10976
Name: count, dtype: Int64

# Seperating independent and dependent variables
df_independent = df_census.iloc[:,:-1]
df_dependent = df_census.iloc[:,-1:]

# dummy columns independent variables
dummy_columns = ['workclass', 'education', 'marital-status', 'occupation', 'relationship', 'race', 'sex', 'native-country']
df_independent = pd.get_dummies(df_independent, columns = dummy_columns, dtype=int)
X = df_independent
X

	age	education-num	capital-gain	capital-loss	hours-per-week	workclass_ Federal-gov	workclass_ Local-gov	workclass_ Private	workclass_ Self-emp-inc	workclass_ Self-emp-not-inc	...	native-country_ Portugal	native-country_ Puerto-Rico	native-country_ Scotland	native-country_ South	native-country_ Taiwan	native-country_ Thailand	native-country_ Trinadad&Tobago	native-country_ United-States	native-country_ Vietnam	native-country_ Yugoslavia
0	39	13	2174	0	40	0	0	0	0	0	...	0	0	0	0	0	0	0	1	0	0
1	50	13	0	0	13	0	0	0	0	1	...	0	0	0	0	0	0	0	1	0	0
2	38	9	0	0	40	0	0	1	0	0	...	0	0	0	0	0	0	0	1	0	0
3	53	7	0	0	40	0	0	1	0	0	...	0	0	0	0	0	0	0	1	0	0
4	28	13	0	0	40	0	0	1	0	0	...	0	0	0	0	0	0	0	0	0	0
...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...
45185	33	13	0	0	40	0	0	1	0	0	...	0	0	0	0	0	0	0	1	0	0
45186	39	13	0	0	36	0	0	1	0	0	...	0	0	0	0	0	0	0	1	0	0
45187	38	13	0	0	50	0	0	1	0	0	...	0	0	0	0	0	0	0	1	0	0
45188	44	13	5455	0	40	0	0	1	0	0	...	0	0	0	0	0	0	0	1	0	0
45189	35	13	0	0	60	0	0	0	1	0	...	0	0	0	0	0	0	0	1	0	0

44962 rows × 102 columns

# re-categorise dependent variables
df_dependent = pd.get_dummies(df_dependent, dtype=int)
y = df_dependent.iloc[:,1]
y

0        0
1        0
2        0
3        0
4        0
        ..
45185    0
45186    0
45187    0
45188    0
45189    1
Name: income_ >50K, Length: 44962, dtype: int64

X_train, X_test, y_train, y_test = train_test_split(X, y, train_size=0.7, random_state=0)

sns.pairplot(df_census, hue="income")

<seaborn.axisgrid.PairGrid at 0x285693890>

2. Decision Tree

from sklearn.tree import DecisionTreeClassifier
tree_clf = DecisionTreeClassifier(max_depth=5)
tree_clf.fit(X_train, y_train)

DecisionTreeClassifier(max_depth=5)

In a Jupyter environment, please rerun this cell to show the HTML representation or trust the notebook.
On GitHub, the HTML representation is unable to render, please try loading this page with nbviewer.org.

from sklearn import tree
fig = plt.figure(figsize=(100, 20), dpi=300)


_ = tree.plot_tree(tree_clf,
                   feature_names=df_independent.columns,
                   class_names=df_dependent.columns, 
                   filled=True,
                   proportion = True,
                   fontsize=18)

y_predict_tree=tree_clf.predict(X_test)

print(classification_report(y_test,y_predict_tree))

              precision    recall  f1-score   support

           0       0.86      0.95      0.90     10099
           1       0.79      0.52      0.63      3390

    accuracy                           0.84     13489
   macro avg       0.82      0.74      0.76     13489
weighted avg       0.84      0.84      0.83     13489

cm = confusion_matrix(y_test, y_predict_tree, labels=tree_clf.classes_)
disp = ConfusionMatrixDisplay(confusion_matrix=cm, display_labels=tree_clf.classes_)
disp = disp.plot()
plt.show()

SVM

from sklearn.svm import SVC

clf = SVC(gamma='auto')
clf.fit(X_train, y_train)
y_predict_SVC = clf.predict(X_test)

y_predict_SVC

array([0, 1, 1, ..., 0, 0, 0])

print(classification_report(y_test, y_predict_SVC))

              precision    recall  f1-score   support

           0       0.87      0.96      0.91     10099
           1       0.82      0.57      0.67      3390

    accuracy                           0.86     13489
   macro avg       0.84      0.76      0.79     13489
weighted avg       0.86      0.86      0.85     13489

cm = confusion_matrix(y_test, y_predict_SVC, labels=clf.classes_)
disp = ConfusionMatrixDisplay(confusion_matrix=cm, display_labels=clf.classes_)
disp = disp.plot()
plt.show()